Date of Award

2024

Document Type

Honors Thesis (Open Access)

Department

Colby College. Psychology Dept.

Advisor(s)

Derek Huffman

Abstract

The present study investigated the structure of spatial representations using a novel computer-based navigation task in concert with computational modeling, specifically with the Successor Representation (SR) model. In particular, we examined whether participant’s knowledge of a virtual environment fit in better with either the cognitive map or labeled graph hypothesis for spatial representation. Twenty-four participants aged 18-22 learned the positions of 25 objects within a 5x5 grid of virtual rooms through disparate routes. Participants then made relative and absolute distance judgments, generated novel shortcuts, and drew a map of the environment to test their knowledge about the spatial locations of each object. Both relative and absolute distance judgments demonstrated systematic biases based on the distance and learned connectivity between the objects, and the SR also generated similar biases when we trained it on our task. The shortcut and map drawing tasks revealed a wide array of individual differences, showing that some participants struggled to integrate spatial knowledge across different routes. Our results from the intersection between computational modeling and the careful study of behavior provide some evidence in favor of the labeled graph hypothesis as well as insights into the large individual variability seen in spatial navigation.

Keywords

navigation, computational modeling, representation, spatial memory

Ainsley Bonin

Available for download on Tuesday, May 20, 2025

Share

COinS